gpt-4o (2024-11-20) is listed as GA but blocked for new deployments — what is going on?

Wang Wang 5 Reputation points
2026-07-01T07:17:16.0766667+00:00

We can't create new deployments of gpt-4o (2024-11-20) or gpt-4o-mini (2024-07-18). Every attempt returns:

ServiceModelDeprecating: The model '...gpt-4o,Version:2024-11-20' is in deprecating state and cannot be used for new deployments.

Here's the problem. Microsoft's own retirement schedule lists both of these versions as GA, retiring October 2026 — nearly 3 months away: https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule?view=foundry-classic

And Microsoft's own lifecycle policy clearly states GA models → "Can create new deployments? Yes." Only the Deprecated stage blocks new customers: https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirements?view=foundry-classic

The same doc maps the API status Deprecating = the Deprecated stage = "blocked for new customers." So the backend is treating these as Deprecated, while the public schedule still says GA. Both cannot be true.

To make it worse — the documented replacement for gpt-4o-mini is gpt-4.1-mini, which is also deprecated and also blocked with the exact same error. So the "replacement" doesn't deploy either.

So I'll ask directly:

  1. Why is a model listed as GA being blocked for new deployments a year before retirement? Is this a bug, or is it silent "new customer limiting" in our region?
  2. If it's intentional — why does the schedule still say GA, and when will it be corrected?
  3. Bottom line: are these models simply not being offered to us anymore? If so, say it plainly instead of showing GA status we can't actually use.

This is a live production workload. We need a straight answer, not a migration script pointing to another deprecated model.

Foundry Models
Foundry Models

A catalog of AI models in Microsoft Foundry that you can discover, compare, and deploy using Azure’s built‑in tools for evaluation, fine‑tuning, and inference

0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Ramin Ahmadi 86 Reputation points
    2026-07-01T21:12:32.3566667+00:00

    Microsoft has moved these models into Deprecated status publicly and the backend is enforcing that state. Existing inference should continue until the stated retirement date, but new deployment creation is no longer reliably available unless the exact subscription/model/version qualifies as “existing customer”. The gpt-4o-mini → gpt-4.1-mini replacement guidance is not sufficient if gpt-4.1-mini is also blocked in your region/deployment type.

    So the practical answer is:

    • If the affected subscription never deployed that exact model/version before, the block is consistent with the Deprecated-stage policy. [learn.microsoft.com]
    • If the affected subscription previously deployed that exact model/version and still cannot create/manage deployments, then it may be a backend eligibility/enforcement issue and should be raised as a support case with that exact evidence.

    Is your subscription classified as an existing customer for that exact model?

    Was this answer helpful?

    0 comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.